Notifications
Clear all

XYZ Calibration Failed. Please consult the manual.  

  RSS
Antimix
(@antimix)
Reputable Member
XYZ Calibration Failed. Please consult the manual.

Hi,

this is a very very weird issue. I have a PRUSA Assembled MK3S+MMU2S.

All was working, until I loaded the latest firmware 3.9. I was told by Prusa support that due to the huge changes in the firmware, it would be better to do a complete FACTORY RESET of the printer just after the flash. So, I have done it, and I discovered that now the printer fails the second setup wizard calibration phase with: XYZ Calibration Failed. Please consult the manual.

I spent the whole week in chat with PRUSA agents, trying everything, with no result.

I run the wizard only one or two time in the life of the printer, just after unboxing it, and it worked, months and months ago. Since then, I never launched the XYZ calibration again.

I am starting to believe that is a general firmware bug.

I tried the test with old firmware 3.7 but the test still failed. However today I remembered that I had 3.5 at the time it worked, so I will test with it.

What really happens to the printer :

I installed a special compiled debug firmware and run the test to have a serial log. What I can see is that during the XYZ calibration it takes a wrong measure of the distances to the referral circles on the bed. And for that reason it assumes is not assembled correctly.

Max. errors (below the maximum allowed error rate)
Max. error X:0.80
Max. error Y:1.50
Max. error euclidian:0.80

measured:
X vector, adjusted: 1.00000, 0.00046
Y vector, adjusted: -0.00039, 1.00000
center, adjusted: -2.54306, 0.25082
Difference after correction:
point #0measured: (10.09940, 6.66004); measured-corrected: (9.45459, 6.25629); target: (12.00000, 6.00000), error: 0.76
point #1measured: (216.80914, 6.98807); measured-corrected: (217.45458, 6.35107); target: (220.00000, 6.00000), error: 0.91
point #2measured: (216.75944, 197.71371); measured-corrected: (217.37937, 198.35105); target: (220.00000, 198.00000), error: 0.89
point #3measured: (10.00000, 197.85289); measured-corrected: (9.37938, 198.25627); target: (12.00000, 198.00000), error: 0.74

Calculate offset and skew returning result:
-2

tmc2130_home_exit tmc2130_sg_homing_axes_mask=0x00
ok
echo:endstops hit: Z:214.08

As you can see, (the exact expected measures are the red one)  is like the X axe is behind 2 mm just at the very beginning, but it looses 4 mm at the end. It is like the TMC2130 is either loosing steps, or a severe build error.  I tend to exclude the build error, because the printer was assembled by PRUSA and never dismounted, never moved, never transported. It was never moved from the table after the initial successfully XYZ calibration one year ago. But there was another firmware.

Another strange thing: issuing

G28 X C

The printer recalibrate the Home-X position and stay there. But if you go to Settings->Move Axis->X Axe, you see that the X axe is at 1.1 mm not zero, and you can not go exactly to zero rotating the wheel from the LCD, and either the motor is at its end-stop, so position zero does not exist.

I want really understand what is happening. I want to be able to identify and solve the issue.
Otherwise the only solution is to ship the printer to PRUSA  and loose a lot of time, since this could require weeks or months.

- Can any of you with a MK3S+MMU2 with firmware 3.9 run the XYZ Calibration test and see if it works ?

Posted : 05/06/2020 11:19 pm
karl-herbert
(@karl-herbert)
Illustrious Member
RE: XYZ Calibration Failed. Please consult the manual.
Posted by: @antimix

Hi,

this is a very very weird issue. I have a PRUSA Assembled MK3S+MMU2S.

All was working, until I loaded the latest firmware 3.9. I was told by Prusa support that due to the huge changes in the firmware, it would be better to do a complete FACTORY RESET of the printer just after the flash. So, I have done it, and I discovered that now the printer fails the second setup wizard calibration phase with: XYZ Calibration Failed. Please consult the manual.

I spent the whole week in chat with PRUSA agents, trying everything, with no result.

I run the wizard only one or two time in the life of the printer, just after unboxing it, and it worked, months and months ago. Since then, I never launched the XYZ calibration again.

I am starting to believe that is a general firmware bug.

I tried the test with old firmware 3.7 but the test still failed. However today I remembered that I had 3.5 at the time it worked, so I will test with it.

What really happens to the printer :

I installed a special compiled debug firmware and run the test to have a serial log. What I can see is that during the XYZ calibration it takes a wrong measure of the distances to the referral circles on the bed. And for that reason it assumes is not assembled correctly.

Max. errors (below the maximum allowed error rate)
Max. error X:0.80
Max. error Y:1.50
Max. error euclidian:0.80

measured:
X vector, adjusted: 1.00000, 0.00046
Y vector, adjusted: -0.00039, 1.00000
center, adjusted: -2.54306, 0.25082
Difference after correction:
point #0measured: (10.09940, 6.66004); measured-corrected: (9.45459, 6.25629); target: (12.00000, 6.00000), error: 0.76
point #1measured: (216.80914, 6.98807); measured-corrected: (217.45458, 6.35107); target: (220.00000, 6.00000), error: 0.91
point #2measured: (216.75944, 197.71371); measured-corrected: (217.37937, 198.35105); target: (220.00000, 198.00000), error: 0.89
point #3measured: (10.00000, 197.85289); measured-corrected: (9.37938, 198.25627); target: (12.00000, 198.00000), error: 0.74

Calculate offset and skew returning result:
-2

tmc2130_home_exit tmc2130_sg_homing_axes_mask=0x00
ok
echo:endstops hit: Z:214.08

As you can see, (the exact expected measures are the red one)  is like the X axe is behind 2 mm just at the very beginning, but it looses 4 mm at the end. It is like the TMC2130 is either loosing steps, or a severe build error.  I tend to exclude the build error, because the printer was assembled by PRUSA and never dismounted, never moved, never transported. It was never moved from the table after the initial successfully XYZ calibration one year ago. But there was another firmware.

Another strange thing: issuing

G28 X C

The printer recalibrate the Home-X position and stay there. But if you go to Settings->Move Axis->X Axe, you see that the X axe is at 1.1 mm not zero, and you can not go exactly to zero rotating the wheel from the LCD, and either the motor is at its end-stop, so position zero does not exist.

I want really understand what is happening. I want to be able to identify and solve the issue.
Otherwise the only solution is to ship the printer to PRUSA  and loose a lot of time, since this could require weeks or months.

- Can any of you with a MK3S+MMU2 with firmware 3.9 run the XYZ Calibration test and see if it works ?

I converted the X / Y axes to 0.9 steppers and had a lot of calibration problems. It was noticed during the calibration that the belt tension exerted a great influence. I also had to adjust the following values in the firmware:

Configuration_prusa.h:

//#define TMC2130_CURRENTS_H {16, 20, 35, 30}  // default holding currents for all axes

//#define TMC2130_CURRENTS_R {16, 20, 35, 30}  // default running currents for all axes

#define TMC2130_CURRENTS_H {18, 22, 35, 30}  // default holding currents for all axes

#define TMC2130_CURRENTS_R {18, 22, 35, 30}  // default running currents for all axes

 

//#define HOMING_FEEDRATE {3000, 3000, 800, 0}  // set the homing speeds (mm/min) // 3000 is also valid for stallGuard homing. Valid range: 2200 - 3000

#define HOMING_FEEDRATE {2600, 2400, 800, 0}  // set the homing speeds (mm/min) // 3000 is also valid for stallGuard homing. Valid range: 2200 – 3000

 

//#define TMC2130_PWM_GRAD_X 2 // PWMCONF
#define TMC2130_PWM_GRAD_X 4 // higher limits current overshoot and avoids squeak (stealthmode), but must be low enough to control accel/decel

//#define TMC2130_PWM_GRAD_Y 2 // PWMCONF
#define TMC2130_PWM_GRAD_Y 4 // higher limits current overshoot and avoids squeak (stealthmode), but must be low enough to control accel/decel

 

//#define TMC2130_SG_THRS_X       3     // stallguard sensitivity for X axis

//#define TMC2130_SG_THRS_Y       3     // stallguard sensitivity for Y axis

#define TMC2130_SG_THRS_X       2     // stallguard sensitivity for X axis

#define TMC2130_SG_THRS_Y       2     // stallguard sensitivity for Y axis

.

I have conducted my experiments with Pronterface. At G28 X C and Y C the values should remain mostly constant. Maybe it would help if you check and change the current values or/and the belt tension a little. It's worth a try. I don't know if Prusa have changed something on the current settings in 3.9.0. You could compare it with older versions.

Statt zu klagen, dass wir nicht alles haben, was wir wollen, sollten wir lieber dankbar sein, dass wir nicht alles bekommen, was wir verdienen.

Posted : 06/06/2020 1:08 am
Antimix
(@antimix)
Reputable Member
Topic starter answered:
RE: XYZ Calibration Failed. Please consult the manual.

@  karl-herbert Interesting points.

Today i flashed to the version 3.5, that was the original firmware the printer had, and I still got the same error in XYZ Calibration.

This is the proof that something happened later to the printer.

So, today I reverted back to the latest 3.9, and oiled again the bearings (of course I did not dismounted the printer, but I just used a syringe to inject oil and WD-40 around the bearings and rods) and then started playing around with axes moving.

What I noticed:

  •  homing is not constant: it is rare to have home X = 0, but most often the min position has a value 1.1, others 1.2. Y Axes at home min. position, go often in negative: -4.0, -2.0
  • I tried to get movements with a calliper, and they seems consistent: 100mm -> 100.02 or 99.98.
  • The PINDA point #0 is:

    point #0 measured: (10.11928, 6.56064); target: (12.00000, 6.00000)

    a small x and y difference (-1.88mm,+0.56mm)  that can caused by the homing error

  • More strange the PINDA point #1:

    point #1 measured: (216.83897, 6.96819); target: (220.00000, 6.00000)

    that has an X error of (-3.16mm, +0.96)  while on the point#0 is -1.88mm.
    If the issue was only on the homing, I would have expected a constant offset on measuring equal in all the measures on an axe.

So, this lead me to think something related to the TMC_2130, that may be have not enough current to perform moves correctly.
Possible causes may be the X and Y axes are too strong to move, due to not good bearings. I measured the required force to move the axes, and they start from 4N with some peak of 6N.

However I did not noticed any artifact on normal prints, and if this would be an issue of the TMC and the movements, I would have expected horrible prints. I have just printed a perfect extrude body part, without any defect.

So ?  🤔 mmmmmh

Posted : 06/06/2020 10:10 pm
karl-herbert
(@karl-herbert)
Illustrious Member
RE: XYZ Calibration Failed. Please consult the manual.

@antimix

I think you have a mechanical problem.

WD40 is not the best joice to oil the bearings.

WD40 is manufactured and marketed as a household lubricant for those "small jobs" or as a rust remover. For these applications, it does a fine job. However, WD40 is not sold nor should it be used as a replacement for the appropriately selected lubricant for rolling element applications.

I would clean the bearings with petroleum or similar, check them for any damage and use a machine oil or soft acid free grease e.g. LM47 or similar.

Statt zu klagen, dass wir nicht alles haben, was wir wollen, sollten wir lieber dankbar sein, dass wir nicht alles bekommen, was wir verdienen.

Posted : 07/06/2020 12:57 am
Share: